Skip to content
Deepak Gupta markguptadeepak.com
By 22 min readGEO

Growth Engineering for the AI Search Era: The Data, Workflow and 10 Claude Code Prompts

Search Console and Bing exports cover half of what you need. The full growth data stack, measurement setup, AI visibility layer and ten Claude Code prompts to build a site that search and AI engines cite.

Growth Engineering for the AI Search Era: The Data, Workflow and 10 Claude Code Prompts, by Deepak Gupta on guptadeepak.com

A question I keep hearing from founders and marketing leaders: is a full export from Google Search Console and Bing Webmaster Tools enough to rebuild a website for search and AI engines? My answer: it is the right starting point, and it covers about half of what you need.

This article is the complete version of that answer. It covers the data stack, the measurement setup, the AI visibility layer, and the exact ten prompts I would give Claude Code to turn raw exports into a prioritized growth plan.

The short version

  • Search Console and Bing exports show how search engines see your site. They do not show which pages produce pipeline, how ChatGPT, Perplexity, Claude or Gemini describe you, or who links to you.
  • The minimum growth data stack has six parts: Search Console, Bing Webmaster Tools (including its AI Performance report), GA4 joined to your CRM, a full site crawl, backlink and mention data, and cross-engine AI visibility. Server logs are a strong seventh.
  • AI referrals are small in volume and large in value. In one Seer Interactive client study, ChatGPT visitors converted at 15.9% against 1.76% for Google organic, on a single site. Measure pipeline by landing page, not sessions.
  • Google says optimizing for its AI features is still SEO. Its May 2026 guide says you do not need llms.txt files, content chunking, AI-specific rewrites or special markup.
  • What earns citations is consistent across studies: original data, pages that are actively maintained, answers stated plainly near the top, and third-party coverage in publications AI engines already trust.
  • The workflow is repeatable: put your exports in one folder, write a CLAUDE.md file with your goals and ICP, connect your AI visibility data through the GrackerAI MCP server, run ten prompts in order, and re-run the core steps every month.

Why the website playbook changed in 2026

Three measurement changes in 2026 made AI search a channel you can actually manage, instead of a guess.

Date Change What it gives you What it still hides
Feb 2026 Bing AI Performance report (public preview; intent, topic and citation share added in June) Citations in Copilot and Bing AI summaries, cited pages, grounding queries Clicks, and any engine outside Microsoft
May 13, 2026 GA4 AI Assistant channel (rolled out globally by early June) AI chatbot referrals grouped as their own default channel Visits that arrive with no referrer, which land in Direct
Jun 3, 2026 Search Console Generative AI reports (all sites since Aug 31) Impressions in AI Overviews, AI Mode and Discover AI features, by page, country, device and date Clicks, CTR, queries, and the AI Overviews vs AI Mode split

At the same time, Google published its first official guide to its AI features on May 15, 2026. The core message: optimizing for generative AI search is still SEO, rooted in the same ranking and quality systems, with a premium on unique, non-commodity content.

Two more shifts matter for anyone who wants to be cited by other sites and the press:

  • AI visitors arrive closer to a decision. A Semrush study measured the average AI search visitor as 4.4 times as valuable as a traditional organic visitor, by conversion rate. Seer Interactive saw ChatGPT at 15.9% conversion vs 1.76% for Google organic on a single site. Treat these as reasons to build the report, not as your forecast.
  • Journalists now research inside AI tools. Muck Rack's State of Journalism 2026 found 82% of journalists use at least one AI tool, with ChatGPT the most used. If an AI answer does not name you, a reporter drafting a story may never find you.

The practical consequence: ranking reports alone no longer describe your visibility. You need to know where you appear, where you are cited, and whether either one turns into revenue.

Is Search Console plus Bing data enough?

No. It is enough to fix your technical layer and understand demand. It is not enough to optimize for leads or for citations, which are the two goals that actually pay.

Here is what the two exports answer well, and where each one goes blind.

Question Search Console + Bing answer it? What fills the gap
Which queries show my pages, and at what CTR and position? Yes Nothing needed
Which URLs are 404, redirected, noindexed or not indexed? Yes, for URLs Google and Bing know about A full crawl finds orphan pages and internal links to broken URLs
Is my structured data valid? Partly: only for rich result types Google reports, on crawled pages A crawl that extracts JSON-LD from every page
Am I cited in Copilot? Yes, through Bing AI Performance Nothing needed
Am I visible in AI Overviews and AI Mode? Impressions only, no clicks or queries AI visibility tracking at the prompt level
Am I mentioned or cited in ChatGPT, Perplexity, Claude or Gemini? No Cross-engine AI visibility (GrackerAI)
Which pages create demo requests, trials and pipeline? No GA4 key events joined to CRM stages
Who links to me, and to my competitors? Barely: the Links report is a sample Ahrefs, Semrush or Majestic
Do AI crawlers reach my important pages? No Server or CDN logs
What do buyers ask that my product does not answer? Partly, in query data Sales notes, support tickets, AI prompt data

One practical warning: the Search Console UI caps performance exports at 1,000 rows. For a real analysis, pull data through the API or the BigQuery bulk export, or you will optimize against a truncated sample.

The complete growth data stack

Every dataset below earns its place by changing a decision. If a dataset would not change what you build, fix or delete, leave it out.

Dataset Typical source Questions it answers Decisions it drives Priority
Search performance, indexing, Generative AI report Google Search Console (API or BigQuery export) Demand, rankings, CTR, index status, AI Overview and AI Mode impressions Retitle, refresh, consolidate, fix indexing Have it
Search performance and AI Performance Bing Webmaster Tools Bing demand, Copilot citations, grounding queries Which pages AI already trusts, and for which questions Have it
Behavior and conversions GA4, joined to CRM (HubSpot, Salesforce) Engagement, key events, AI Assistant traffic, pipeline by landing page Where to invest by revenue, not traffic Add first
Full site crawl Screaming Frog, Sitebulb Every URL, status code, canonical, internal link, schema block, word count, orphan page Technical fixes Search Console cannot see Add first
Backlinks and brand mentions Ahrefs, Semrush, Majestic Referring domains, linked pages, competitor link magnets, unlinked mentions Digital PR targets and which assets to build Add first
Cross-engine AI visibility GrackerAI Mentions, citations, share of voice, sentiment, claim accuracy and prompt gaps across AI engines Which prompts to win and which sources to influence Add first
Server or CDN logs Cloudflare, Fastly, your host Which bots fetch what, how often, and which errors they hit AI crawler access, crawl waste Recommended
ICP and competitor brief You, written once Who buys, why, and who you lose to How every gap gets scored Required, takes an hour
Voice of customer Sales call notes, support tickets, churn reasons Objections, missing features, jobs to be done Product roadmap and bottom-of-funnel content Recommended

If you can only add one dataset this week, add GA4 joined to CRM outcomes. Without it, every other analysis optimizes for traffic, and traffic is not the goal.

Setting up measurement before you optimize

Spend one day on measurement before you touch content. Every later decision depends on it, and AI traffic is easy to undercount.

1. GA4: use the AI Assistant channel, then back it up

GA4 added a native AI Assistant channel on May 13, 2026. It assigns the medium ai-assistant when the referrer matches a list Google maintains, and it reached all properties by early June. It does not backfill history, and it cannot see visits that arrive with no referrer, which still show up as Direct.

Add a custom channel group as a backup so new assistants are counted the day they send traffic. Use a source regex like this one and extend it as engines change:

(chatgpt\.com|chat\.openai\.com|perplexity\.ai|claude\.ai|gemini\.google\.com|copilot\.microsoft\.com|grok\.com|deepseek\.com|meta\.ai|you\.com)

Then do three things most teams skip:

  1. Mark demo requests, trial signups, contact submissions and newsletter signups as key events.
  2. Pass the first landing page and channel into your CRM, so pipeline can be attributed to a page.
  3. Add "AI assistant (ChatGPT, Perplexity, etc.)" as an option on your "How did you hear about us?" field. Self-reported attribution catches the AI influence your analytics will miss.

2. Search Console: export through the API

Pull 16 months of performance data by page and query through the API or BigQuery bulk export. Add the page indexing report, the sitemaps report, and the Generative AI report. Remember the Generative AI report is impressions only: clicks from AI features stay in the main Performance report, mixed in with ordinary web results.

3. Bing Webmaster Tools: export AI Performance

The AI Performance report shows citations and cited pages over time, plus the grounding queries Copilot used to find you. Since June 2026 each grounding query also carries an intent, a topic and your citation share. It covers Copilot, Bing AI summaries and some partner integrations, with no click data. Treat the grounding queries as a free list of questions AI already associates with your pages.

4. Logs: confirm AI crawlers reach you

Export 30 days of CDN or server logs and filter by user agent: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-User, Claude-SearchBot, PerplexityBot, Perplexity-User, Googlebot and Bingbot. You are looking for important pages that never get fetched, and for 4xx or 5xx errors served to bots.

Be careful with blocking. A Rutgers and Wharton study found news publishers that blocked AI crawlers in robots.txt saw about 7% fewer weekly visits within six weeks. Block training if you must, but think hard before blocking the bots that power live answers.

AI visibility across engines with the GrackerAI MCP

Search Console and Bing tell you about Google and Microsoft surfaces. They say nothing about ChatGPT, Perplexity, Claude or Gemini, which is where a growing share of B2B research now starts. That is the gap GrackerAI, my company, covers, and its MCP server is the fastest way to bring that data into the same Claude Code session as your exports. Any cross-engine tracker with an export or an API can fill the same slot in this workflow.

What the MCP server exposes

The GrackerAI MCP server gives any MCP-compatible assistant live, read-only access to your visibility data:

  • Visibility: scores, presence and mention counts by prompt, engine and country
  • Mentions: placement, prominence, sentiment and claim accuracy
  • Citations: cited URLs, domain, source type, publish date, and whether the source supported or contradicted a claim
  • Competitors: share of voice and the brands engines recommend instead of you
  • Prompts: the tracked prompt set with intent and funnel stage

Monitoring covers up to ten engines: ChatGPT, Perplexity, Google AI Mode, Google AI Overviews, Gemini, Microsoft Copilot, DeepSeek, Grok, Claude and Brave Leo, depending on the plan.

Setup takes about 30 seconds

It is a hosted server. There is nothing to install and no API key to paste, and every query is read-only: nothing in your workspace can be created, edited or deleted through it. Each request counts against your plan's monthly MCP limit.

Claude Code:

claude mcp add --transport http gracker https://portal.gracker.ai/mcp

Then run /mcp inside Claude Code, choose gracker, and sign in with your GrackerAI account in the browser window that opens.

Claude (web or desktop): Settings, then Connectors, then Add custom connector. Paste https://portal.gracker.ai/mcp and authorize.

ChatGPT, Copilot, Perplexity and Grok: Settings (or Connectors), then Add custom connector, and paste the same URL. Cursor and VS Code: add it as an HTTP server in .cursor/mcp.json or .vscode/mcp.json.

During sign-in you choose which brands the connection can see. Access never exceeds your own account permissions, and removing the connector from your client cuts it off.

Questions worth asking it

  • "What is my visibility score for the last 7 days, by engine?"
  • "Which prompts do competitors appear in where I do not?"
  • "Which domains cite my competitors but never cite me?"
  • "Show every mention with negative sentiment or an inaccurate claim in the last 30 days."
  • "Which of my URLs get cited most, and for which funnel stage?"
  • "Give me a weekly report: share of voice change, new citation sources, lost prompts."

Why this matters in the workflow

The value is the join. In one session, Claude Code can read your Search Console export, your GA4 conversions and your live AI citation data, then answer questions none of them answers alone. For example: which pages rank on page one, convert well, and are still never cited by ChatGPT? Those are your highest-leverage rewrites.

Preparing the Claude Code workspace

The quality of Claude Code's output depends far more on the workspace than on clever prompting. Give it clean folders, a written brief, and permission to write scripts instead of eyeballing data.

Folder layout

growth/
  CLAUDE.md                 # goals, ICP, rules (template below)
  data/
    raw/
      gsc/                  # performance by page+query, indexing, sitemaps, generative AI
      bing/                 # search performance, AI performance, crawl issues
      ga4/                  # landing pages x key events, AI Assistant channel
      crm/                  # leads and opportunities with first landing page
      crawl/                # Screaming Frog internal_all, schema, inlinks
      backlinks/            # referring domains, top linked pages, competitor links
      logs/                 # 30 days of bot requests
    clean/                  # normalized tables Claude Code writes
  scripts/                  # every analysis as a re-runnable script
  outputs/                  # CSVs and briefs, one folder per step

Connect the GrackerAI MCP once (previous section), and AI visibility data stays live instead of becoming another stale export.

CLAUDE.md template

Claude Code reads this file at the start of every session. Fill it in once; it keeps every later prompt grounded in your business, not generic SEO advice.

# Growth project: <company>

## Business
- Product: <one sentence, what it does and for whom>
- ICP: <role, company size, industry, trigger event>
- Conversion events, in order of value: demo request > trial > contact > newsletter
- Average deal size and sales cycle: <numbers>

## Site
- Domain: <domain>, CMS/framework: <stack>
- Money pages: /pricing, /demo, /product/*, /integrations/*
- Content hubs: /blog, /guides, /research

## Competitors
- <competitor 1>, <competitor 2>, <competitor 3>
- Publications our buyers and journalists read: <list>

## Data
- Raw exports in data/raw (see folder names). Date range: <range>
- AI visibility: GrackerAI MCP (live, read-only)

## Rules
- Write a script for every analysis. Never estimate numbers by reading files.
- Normalize URLs before joining: lowercase host, strip tracking params, consistent trailing slash.
- Rank opportunities by lead potential, then effort. Traffic alone is not a reason.
- Plan first, wait for my approval, then execute.
- Save outputs as CSV in outputs/<step>/ with a short README of findings.
- Do not change the live site. Propose changes as diffs or briefs.

The 10 Claude Code prompts, in order

Run these one at a time, in a single project, so each step builds on files the last one wrote. Review the plan before approving each step. The first run takes a few days of review; later runs take hours.

Step 1. Ingest and normalize

Read every file in data/raw. Build a data dictionary in outputs/01-ingest/README.md: source, columns, date range, row count, and known limits (sampling, row caps, impressions-only reports). Normalize URLs per CLAUDE.md and build one page-level table in data/clean/pages.csv joining GSC, Bing, GA4, CRM, crawl and backlink data. Report join rates and every URL that failed to match. List missing or inconsistent data before any analysis. Write all of it as scripts in scripts/.

You get: a single source of truth per URL, and an honest list of data gaps before they distort later steps.

Step 2. Technical health and 404 recovery

Using GSC and Bing indexing data, the crawl, and logs, list every 404, soft 404, redirect chain, redirect loop, canonical conflict, noindexed page in the sitemap, and "crawled, currently not indexed" URL. For each 404 that has impressions, backlinks, or internal links pointing to it, propose a 301 target with a one-line reason. Flag important pages that AI crawlers never fetched. Output redirects.csv, internal-link-fixes.csv, and a fix list ranked by impact x effort.

You get: a redirect map that recovers link equity, and a list of indexing problems ranked by what they cost you.

Step 3. Query to page mapping

Map every query to the pages that receive its impressions. Flag cannibalization (two or more URLs splitting one intent), striking-distance queries (position 5 to 20 with real impressions), and pages whose CTR is well below the site median for their position band. For the top 30 opportunities by lead potential, draft new titles and meta descriptions, and say which page should own each cannibalized query.

You get: quick wins that need copy changes, not new content.

Step 4. Content inventory and decisions

Score every indexable page on: 12-month click trend, conversions and pipeline (GA4 + CRM), AI citations (Bing AI Performance + GrackerAI MCP), referring domains, and last-updated date. Classify each page as keep, refresh, merge, redirect, or remove, with the reason and the target URL for merges. Flag pages older than 12 months that still earn citations or leads: these are refresh priorities. Output content_decisions.csv.

You get: a keep, refresh, merge or remove call on every URL, defended with data.

Step 5. Content and prompt gaps

Cluster all GSC queries, Bing grounding queries, and GrackerAI tracked prompts into topics. Tag each topic by buyer stage: problem, solution, comparison, implementation. Using the GrackerAI MCP, find prompts where competitors are mentioned or cited and we are not, and list the domains cited in those answers. Show topics with demand and no matching page, and missing page types: comparisons, alternatives, integrations, use cases, pricing explainers. Rank by lead potential, then effort. Write a one-paragraph brief for the top 15.

You get: a content roadmap built from real demand in both search and AI answers, ranked by revenue potential.

Step 6. Structured data and entity consistency

Extract JSON-LD from every crawled page and group by template. Report missing, invalid, or inconsistent schema. Generate valid JSON-LD per template: Organization (with sameAs to every official profile), SoftwareApplication or Product, Article with author as Person, BreadcrumbList, and HowTo or FAQPage only where the content truly matches. Check that the company name, product name, founders, and description are identical across schema, the about page, and official profiles. Output one JSON-LD file per template and a validation report.

You get: clean, consistent entity data that search engines and AI systems can trust.

Step 7. AI answer readiness

For the top 50 pages by leads and impressions, check: is the main answer stated in the first 100 words; does each section stand on its own when read in isolation; are claims backed by a number, a source, or first-hand experience; is there a named author with credentials; is the updated date visible and accurate; is the page server-rendered and reachable by AI crawlers. Using the GrackerAI MCP, compare what AI engines say about us against what the pages say, and flag inaccurate claims. Rewrite the intro and one key section for the top 10 pages as examples. Do not add filler or keyword stuffing.

You get: pages written so a person and a retrieval system both get the answer fast, plus a list of wrong claims AI engines make about you.

Step 8. Citation-worthy assets

From the backlink data, find the competitor and industry pages that earn the most referring domains, and classify what they are (research, statistics, tools, templates, glossaries). Using the GrackerAI MCP, list the source types AI engines cite most for our priority prompts. Propose 5 assets we can own: an annual benchmark report from our product data, a sourced statistics page, a methodology page, a glossary, and a free tool. For each, give the outline, the data we need, the update cadence, and 20 publications or writers likely to cite it.

You get: a short list of assets designed to be cited by journalists, analysts and AI engines, with the outreach list attached.

Step 9. Conversion paths

For every page with meaningful traffic, check whether the call to action matches intent: education pages should offer a next read or a resource, comparison and pricing pages should offer a demo or trial. List pages with traffic and zero conversions, pages missing internal links to money pages, and AI-referred landing pages with high exits. Draft specific fixes: CTA copy, placement, internal links, and one lead magnet per topic cluster.

You get: more pipeline from the traffic you already have, which is usually the cheapest growth available.

Step 10. Product insights and the monthly loop

Mine queries, grounding queries, AI prompts, and sales and support notes for: feature requests, integrations people search for, objections, use cases we do not serve, and competitor frustrations ("X alternative", "X too expensive"). Summarize them as a product input memo. Then build scripts/monthly.py that re-runs steps 2 to 5 on fresh exports, diffs the results against last month, and writes a scorecard: clicks, leads by landing page, AI citations and share of voice, referring domains, indexing errors.

You get: a product memo grounded in market demand, and a growth system that runs every month instead of a one-time audit.

What earns citations from AI engines and journalists

The goal of being cited "automatically" by other sites, news and AI answers is realistic, but nothing about it is automatic. You earn it by being the original source of something people need to reference.

Five findings from recent research point the same way:

Finding Evidence What to do
Maintained pages win 75% of pages cited by ChatGPT, Gemini and Perplexity were updated in the last year, and more than a quarter of those were first published over two years ago (Seer Interactive, July 2026) Refresh your best pages on a schedule instead of only publishing new ones
Original data wins Cision's survey of 1,899 journalists lists original research and credible data among what makes a pitch work, and 66% of journalists rely on PR for story ideas (Cision 2026) Publish numbers only you have, with sample size and method next to the claim
Off-site authority still counts Sites with over 32,000 referring domains were 3.5 times more likely to be cited by ChatGPT than sites with up to 200 (SE Ranking) Treat digital PR as part of AI visibility, not a separate program
Earned media feeds AI answers Earned media made up 84% of citations, and journalism alone about 27%, in Muck Rack's analysis of more than 25 million links cited by ChatGPT, Claude and Gemini (Muck Rack, May 2026) Get covered in the publications AI engines already cite for your category
Journalists use AI, and want sources 82% of journalists use at least one AI tool (Muck Rack 2026) Pitch one striking number, the method, and an expert, in a short note

The citation asset playbook

These formats get referenced repeatedly because writers need them while drafting:

  1. An annual benchmark report built from your own product or customer data. This is the hardest for competitors to copy.
  2. A sourced statistics page for your category, with every number linked to its original source and refreshed quarterly. I keep one for AI search at The State of AI Search in 2026.
  3. A methodology page explaining how your data was collected, so journalists and AI systems can trust and attribute it.
  4. A glossary of category terms with plain definitions, which AI engines and writers use for "what is" questions.
  5. A free tool or calculator that produces a result people screenshot and share.
  6. Honest comparison pages that name competitors, show real tradeoffs, and stay current.

Make each asset easy to cite: a clear title with the year, a named author with credentials, a visible updated date, charts with a source line burned into the image, and a suggested citation at the bottom. Add a newsroom page with your data, logos, founder bios and a contact who answers within a day.

What to skip

A lot of AI search advice in circulation has never been tested. Here is what the evidence says you can drop, or at least stop paying for.

  • llms.txt as a visibility lever. Ahrefs found 97% of valid llms.txt files received zero requests of any kind in May 2026, and Google's guide says Google Search ignores them. It costs little to keep one, but it will not move citations.
  • Chunking and AI-specific rewrites. Google's May 2026 guide lists content chunking, rewriting for AI systems and special markup among the things you can ignore. Clear writing helps people and machines alike; artificial fragmentation helps neither.
  • Schema as a magic switch. Structured data clarifies entities and enables rich results. In Ahrefs' matched study of 1,885 pages, adding JSON-LD produced no meaningful citation lift. Fix the content first.
  • Manufactured mentions. Google's guide lists seeking inauthentic "mentions" among the tactics to avoid. Planted forum posts and paid listicle placements are a short-term trick with long-term costs.
  • Blanket AI crawler blocks. Decide bot by bot. Blocking the agents that fetch pages for live answers removes you from those answers.
  • Rank tracking as the only KPI. Position one with an AI Overview above it is a different asset than position one in 2023. Track citations and pipeline next to rankings.

For product teams: search data is a roadmap input

Search and AI prompt data are the largest unprompted survey of your market you will ever get. People type what they actually want, in their own words, with no sales rep in the room.

Signal in the data What it usually means Product or GTM response
"<your product> + <tool> integration" queries with no page Buyers need the integration to adopt Build it, or document the workaround, then publish the integration page
"<competitor> alternative" and "<competitor> too expensive" A competitor is losing buyers on price or fit Honest comparison page, migration guide, pricing tier aimed at switchers
"How to <task> in <your product>" with high exits The feature exists but is hard to find or use Fix the UX first, then the docs
"<your product> pricing" with low CTR or high bounce Pricing is unclear or hidden Publish clear pricing, or explain why it is custom
AI engines state wrong facts about you Your own pages are vague, stale or contradicted elsewhere Publish a product facts page and correct the sources AI cites
Prompts where AI recommends a competitor for a use case you serve You have no proof for that use case Ship a case study, a template and a dedicated use-case page

Claim accuracy deserves special attention. If ChatGPT tells buyers your product lacks SSO or costs twice what it does, that error shows up in sales calls before it shows up in any dashboard. The GrackerAI MCP flags inaccurate claims, and Step 7 of the workflow ties each one to the page that should correct it.

The best SaaS sites treat documentation, changelogs, API references and templates as growth assets. They answer implementation questions, which are among the highest-intent questions a buyer asks.

Scorecard and operating cadence

A growth program is a loop, not an audit. Track a small scorecard and run the same steps on a fixed rhythm.

The scorecard

Metric Source Why it matters
Pipeline and leads by landing page GA4 + CRM The outcome every other metric serves
AI Assistant sessions and their conversion rate GA4 (native channel + custom group) Value of AI visibility, in revenue terms
Share of voice on priority prompts, by engine GrackerAI Whether AI engines recommend you or a competitor
Citations and cited pages Bing AI Performance, GrackerAI Which pages AI systems trust
AI Overview and AI Mode impressions Search Console Generative AI report Exposure in Google's AI surfaces
Non-brand clicks and CTR Search Console Classic search demand captured
New referring domains and unlinked mentions Ahrefs or Semrush Whether citation assets are working
Indexing errors and 404s with value Search Console, crawl Technical leakage
Inaccurate AI claims open GrackerAI Brand and sales risk

The cadence

Rhythm What to do Workflow steps
Weekly (30 minutes) Check new indexing errors, AI citation and share-of-voice changes, and new negative or inaccurate mentions MCP questions, Step 2 alerts
Monthly Pull fresh exports, run the monthly script, refresh 5 to 10 pages, ship 2 to 4 gap pages, update the statistics page Steps 2 to 5, 7, 9
Quarterly Publish one original research asset, run a digital PR push, full crawl and schema review, revisit ICP and competitors Steps 6, 8, 10
Yearly Annual benchmark report, prune or merge content that no longer earns traffic, leads or citations Steps 4, 8

Closing thoughts

The teams winning AI search in 2026 are not doing anything exotic. They measure what converts, fix what is broken, keep their best pages current, and publish data that other people need to quote.

What changed is the speed. With clean exports, a CLAUDE.md file and live AI visibility data through an MCP server, one marketer and one product lead can run in a week the analysis that used to take an agency a quarter.

Start with measurement. Run the ten steps. Then do it again next month.

That is growth engineering.

Further reading

Sources

More like this

All GEO & AI Search

From another track

Get new GEO & AI Search writing

Enjoyed this? Subscribe and tell us what you read most. GEO & AI Search is already ticked for you. No tracking pixels, unsubscribe with one click.

Tell us what you read most (optional)

About DeepakPublicationsAnalysisAll tracks